On Mon, Jun 11, 2007 at 07:09:19PM -0500, James Graves wrote: > Greg KH wrote: > > On Tue, Jun 05, 2007 at 03:26:07PM -0500, James Graves wrote: > >> Hello all, > >> > >> We're seeing a strange problem when writing commands to a Sierra Wireless > >> USB modem. It is a USB 1.1 device. > >> > >> We're sending some commands over the control channel for the modem. Almost > >> everything works fine. > > How are you sending these commands? The current in-kernel driver > > doesn't support this from what I can see, right? > > This is from a user-space application. There's no special support in > the in-kernel driver, nor does there need to be.
Ok, good, I was confused there for a bit :) > >> However, when sending commands larger than 64 bytes, the rest of the data > >> (which was a part of that write) is corrupted, which we've seen using a > >> USB sniffer. > > Corrupted where, coming from the host to the device? > > This is coming from the host, going to the modem. What host controller are you using? Is it the built-in ohci controller that is in many of these chipsets or are you using a development board with a "raw" usb connection? > > Are you sure you aren't just forgetting to "encode" your data properly > > as the device specs show that you need to do so? I can dig up the > > relevant reference if you need it. > > Yeah, pretty sure. In the application, we're dumping out the data right > before the glib call to write the data to the modem, and it is in the > proper format with the correct escape sequences as needed. We can add a small change to the driver to dump out what it thinks it is sending to the device in case it is different. Let me know if you want that functionality added. > >> We have modified the userspace application so that it writes only up to 64 > >> bytes at a time, sleeps a little, and then writes more. This works fine. > > This sounds odd, what endpoint pair are you sending to? > > This is endpoint 4 (bulk), which is using Sierra Wireless's control > protocol. > > >> The strange part is that we've seen no issues with using the PPPD > >> application to connect to the Internet using the Sierra Wireless modem. > >> And it has got to be writing more than 64 bytes at a time, though that is > >> to a different device file (and therefore a different bulk endpoint). > > Which device is this? > > Sorry, should have mentioned this to begin with. It is the MC8775. > > We've been doing some further testing, and there are some cases where > everything seems to work fine. > > So we haven't yet completely isolated the conditions where things break > or work correctly. It may well not be a USB transport issue at all. > > As suggested, we've also been looking at the output of usbmon. However, > there are some things happening there that I don't quite understand, which I > can post about separately. That would be good to hear about. thanks, greg k-h ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
